Venture capital has always thrived on growth narratives, but the rise of AI has pushed the boundaries of financial reporting to their limits. Simon Glairy, a distinguished expert in risk management and AI-driven assessment, provides a sobering look at how the industry is currently navigating the “ARR scam.” With the pressure to scale from $1 million to $100 million in record time, many startups are blurring the lines between actual revenue and future promises. This conversation explores the shift toward “committed” metrics, the risks of usage-based billing, and the long-term consequences of prioritizing public relations over transparent bookkeeping.
Startups often group “committed but not yet live” contract values into their primary revenue metrics. How does this reporting style impact internal resource allocation, and what specific safeguards should a finance team implement to prevent these figures from diverging too far from actual cash flow?
The primary danger of treating Committed ARR (CARR) as realized revenue is that it creates a false sense of security that can lead to aggressive over-hiring or premature scaling. When a finance team sees a high top-line figure, they might greenlight an $8 million expansion plan based on money that hasn’t actually hit the bank account yet. To safeguard against this, I recommend a “implementation-gated” budget where only a percentage of committed funds are unlocked for spending based on technical milestones. Finance teams must maintain a rigid distinction in internal dashboards between ARR and CARR, ensuring that cash runway calculations are strictly tied to GAAP-recognized revenue rather than future promises. By requiring a 20% to 30% “liquidity buffer” on all committed contracts, firms can avoid the sensory shock of a sudden cancellation during a long deployment phase.
Large-scale free pilots and heavily discounted multi-year contracts are frequently treated as guaranteed future income. What specific criteria should a founder use to discount these contracts for churn risk, and how do you manage board expectations when implementation delays occur?
Founders often fall into the trap of counting a one-year free pilot as high-intent ARR, but history shows these are some of the most volatile “contracts” in existence. A founder should apply a historical churn multiplier—often as high as 50% for unproven AI integrations—to any pilot that hasn’t reached a “paid” status. Managing the board requires absolute transparency regarding the “time-to-value” metric; if an implementation takes six months instead of two, the board needs to hear about the technical friction immediately, not at the end of the quarter. It is vital to report “downsell risk” alongside growth figures, providing a clear-eyed view of how many customers might opt out before reaching the higher-priced years of a discounted multi-year deal.
Investors often benefit from aggressive public revenue declarations that position a portfolio company as a category leader. What are the long-term trade-offs of using these metrics to attract top-tier talent, and how does this strategy complicate the due diligence process during future funding rounds?
While a $100 million ARR headline can certainly lure elite engineers, it creates a “valuation trap” where the internal reality cannot sustain the external hype. When those employees realize the actual revenue is closer to $40 million, morale craters and the “fake it till you make it” culture begins to erode institutional trust. During future funding rounds, sophisticated investors will dig into the “remaining performance obligations” and quickly spot the $8 million to $10 million gaps that were previously dismissed as rounding errors. This discrepancy complicates due diligence because it forces the founder to defend a narrative rather than a business model, often leading to punishing “down rounds” or flat valuations that wipe out employee equity.
AI companies increasingly rely on usage-based or outcome-driven billing rather than fixed subscriptions. When extrapolating a single week or month of high usage into an annual figure, what volatility risks emerge, and what alternative metrics provide a more stable picture of company health?
The risk of “annualizing” a peak week of AI usage is immense; a single customer’s heavy testing phase can artificially inflate an annual run-rate by millions of dollars. If a startup has one massive month of inference costs and billing, extrapolating that by 12 creates a fragile financial house of cards that collapses the moment usage stabilizes. Instead of just looking at annualized run-rate, founders should focus on “Net Revenue Retention” and “Daily Active Usage” trends to see if the growth is structural or just a spike. Measuring the “cost-to-serve” alongside usage-based revenue is also critical to ensure that high usage isn’t actually burning through margins, which is a common occurrence in compute-heavy AI applications.
High revenue figures can lead to exceptionally high valuation multiples that may become difficult to justify during market corrections. How can founders balance the need for rapid growth narratives with the necessity of transparent bookkeeping, and what steps prevent a company from falling into a valuation trap?
The most effective way to balance growth and transparency is to adopt a “dual-track” reporting system: use aggressive growth metrics like CARR for marketing and talent acquisition, but use conservative GAAP metrics for all board-level and strategic decision-making. Founders must resist the urge to believe their own PR; just because an investor is “looking the other way” during a bull market doesn’t mean they won’t use those same inflated numbers against you when the market turns. To avoid the trap, founders should set their internal valuation targets based on 10x or 15x of actual realized ARR rather than the 50x to 100x multiples often seen in hype cycles. This grounded approach ensures that even if a correction occurs, the company has enough “meat on the bone” to justify its existence to the next set of auditors.
What is your forecast for AI startup revenue reporting?
I expect we are heading toward a “Great Reconciliation” where the gap between public declarations and audited books will finally be forced shut by late-stage investors and public market standards. As the initial AI hype settles, the industry will move away from “annualized run-rates” based on single-week spikes and return to a more disciplined focus on “Remaining Performance Obligations” (RPO) and actual cash collections. We will likely see more founders being called out on social media and in the press for “CARR-washing,” leading to a new standard of “Clean ARR” that excludes pilots, discounts, and un-onboarded contracts. Ultimately, the startups that survive the next three years will be those that prioritize high-quality, predictable revenue over the short-term dopamine hit of a misleading press release.
